A class-D amplifier or switching amplifier is an electronic amplifier in which the amplifying devices (which are transistors, usually field effect transistors, i.e., FETs) operate as electronic switches, and not as linear gain devices as in other amplifiers. Such amplifiers are well known to those of skill in the art as an architecture that can achieve high amplification with high efficiency.
The signal to be amplified is a train of constant amplitude pulses, so the active devices switch rapidly back and forth between their fully conductive and nonconductive states. The analog signal to be amplified is converted to a series of digital pulses by pulse width modulation, pulse density modulation or other method before being applied to the amplifier.
After amplification, the output pulse train can be converted back to an analog signal by passing it through a passive low pass filter consisting of inductors and capacitors. (In some applications the low pass filter is omitted and the inherent bandwidth of the transducer, such as a loudspeaker, functions as the low pass filter.) The major advantage of a class-D amplifier is that it is typically more efficient than comparable analog amplifiers, with less power dissipated as heat in the active devices.
Output stages such as those used in pulse generators are examples of class-D amplifiers. However, the term mostly applies to power amplifiers intended to reproduce signals with a bandwidth well below the switching frequency.
FIG. 1 is an illustration of a class-D amplifier 100 in a basic single-ended output, or half bridge, design. The input signal 102 is combined with a triangular wave signal to produce a train of square pulses 104 of fixed amplitude but varying width and separation; the low-frequency portion of the signal is the signal to be amplified, while the high-frequency portion makes the waveform binary.
The pulse train 104 is amplified by the output stage 106, resulting in a pulse train 108 having the same frequency spectrum, but with greater amplitude. The amplified pulse train 108 is then converted back to an analog signal by passing it through low pass filter 110, which removes the unwanted high-frequency components, and the analog signal then passed to loudspeaker (or other load) 112. The low-pass filter 110 is of high efficiency due to the use of only reactive components, such as inductors and capacitors.
In some applications, particularly those used in portable mobile devices, the inductor which is part of the low-pass filter 110, and which is in series with the loudspeaker load, may be omitted. This is because the inductance of the loudspeaker voice coil may be enough to perform the filtering of the switching waveform needed to remove the high frequency components of the driving waveform.
While the load is illustrated as a loudspeaker in FIG. 1 and other figures herein, one of skill in the art will appreciate that non-audio applications, and loads other than loudspeakers, can benefit from the invention described herein.
As is known in the art, the output power of this amplifier is limited by the value of the supply voltage VDD used at the output stage. In many applications of a class-D amplifier, the supply voltage is obtained from a battery. For example, portable mobile devices are often powered by lithium batteries that may have a primary voltage of around 4.2 volts (V) when fully charged, falling to perhaps 3.2V as the battery is discharged. As a result, the output power that can be delivered into a loudspeaker load having, for example, a resistance value that might typically be 8 Ohms is limited to only a few hundred milliwatts or less.
In one attempt to overcome this limitation, the class-D amplifier circuit is often modified to create a second, complimentary output so that the loudspeaker load is now driven from both ends in opposite phases. This is well known as “balanced” or “differential” driving; the output stage is typically referred to as a “full bridge” or “H-bridge” circuit, and the loudspeaker or other load is referred to as a “bridge-tied” load (“BTL”). A controller (not shown) provides the inputs to the two sides of the bridge in opposite phases.
FIG. 2 is an illustration of a differential class-D amplifier output stage 200 using a full bridge circuit. The output bridge is comprised of FET transistors MHA, MLA, MHB, and MLB, to which the input signal is fed by a controller (not shown). In this configuration the output voltage swing has doubled as there is now a second complementary output so that the loudspeaker load is now driven from both ends in opposite phases. Since power is proportional to the square of the voltage across the load, the power delivered to the loudspeaker is nominally increased by four. As is known in the art, for the loudspeaker load to be driven in a differential fashion, each side of the bridge must be a mirror image of the other, and thus able to provide the same voltages, in opposite phases as above.
In the circuit of FIG. 2, as is typical in the art, transistors MHA and MHB are p-channel metal oxide semiconductor (PMOS) transistors, while transistors MLA and MLB are n-channel metal oxide semiconductor (NMOS) transistors, although this is not necessary. Those of skill in the art will appreciate the difficulties involved in using NMOS transistors rather than PMOS transistors and vice versa. Because PMOS devices have a lower conductivity than NMOS devices for any given physical size, it is not uncommon to use a bridge where all the devices are NMOS devices, even though this makes the drive waveforms more difficult to generate, because the NMOS devices that are substituted for the PMOS devices require a gate drive above the maximum supply to the bridge.
However, even the increase in power from circuit 200 will only produce just over a watt of average power in a mobile device from a 3.6 volt supply voltage into an 8 ohm load (3.6×3.6×⅛×0.707=1.134 watts). In an automotive application with a 12 volt supply voltage and a 4 ohm load, the average power would be about 25 watts (12×12×¼×0.707=25.4 watts).
At these maximum power levels the control signals to the output bridge will tend towards a 100% duty cycle. That is, the bridge is almost permanently on one way, steering current into the load. In this situation the amplifier has become ‘saturated’ and cannot produce more power in response to a larger input signal. The bridge thus can no longer behave in a linear fashion, and the signal becomes ‘clipped’ at its maximum level, a form of heavy distortion. As a result, it is desirable to design an amplifier in a way that avoids this ‘clipping’ condition, although some designs have the effect of further reducing the output power.
In order to increase these power levels, the signal swings across the loudspeaker load must be increased. One typically way of achieving this is by ‘boosting’ the supply voltage to the full bridge output stage. In some cases, this is done by using standard supply circuit design techniques such as switch mode boost converter circuits.
FIG. 3 is a block diagram of one commercial implementation of a class-D amplifier 300 that uses such a boost converter combined with a full bridge output amplifier. This form of supply voltage boosting uses a discrete inductor component LX and the design of a controller (“CONTROL”) 302. The use of a boost converter 304 results in a supply voltage VCCOUT that might, for example, be 9 to 12 volts rather than the nominal battery voltage of about 4 volts. The increased voltage VCCOUT goes to the output stage 306 to increase the power to the loudspeaker load.
The controller 302 looks at the input signal and determines whether the input signal is such that a boost of supply voltage is needed; if a boost is appropriate, the boost converter 304 is activated. However, it is desirable that the boost converter 304 not run all the time, as this will drain the battery faster than when the boost converter 304 is not active. Thus, to optimize efficiency of the amplifier 300, the controller 302 only activates the boost converter 304 when the desired output signal of the amplifier becomes large enough to require the increase in supply voltage that the boost converter 304 provides. In some embodiments, the supply voltage can be made to “track” the signal level, further increasing efficiency of the amplifier 300, although at the cost of increased circuit complexity.
Further, boost converters are inconvenient, both due to the additional cost of inductors over capacitors, and, more importantly, the physical size of inductors. The height of an inductor on a printed circuit board may be 2 to 3 millimeters (mm), while a capacitor is less than 1 mm high. This places a limitation on the design of, for example, a smartphone. Finally, to limit the size of the inductor, it is desirable to drive it at a high frequency, such as 1 megahertz (Mhz), which can cause electromagnetic interference.
Another possible solution is a charge pump, a device well known in the art that is frequently used to create larger or smaller supply voltages for many applications. Charge pumps create continuous supply voltages that can be used to power circuits in a manner similar to the inductor based boost circuit above. However, charge pumps with physically small capacitors must run at high speeds, which reduces their efficiency. Also, as opposed to inductive boost circuits, charge pumps are unable to efficiently provide a continuous voltage and thus cannot efficiently track the optimum supply level.
An example of such a charge pump implementation is seen in the TDA1560 audio amplifier from NXP Semiconductors N.V. (previously Philips Semiconductors), which uses a large electrolytic capacitor to lift the supply voltage during large audio signal peaks, thus achieving larger peak output power levels. In this case the maximum output power and minimum signal frequency are functions of the capacitor value that is chosen in the range 2200 microfarads or larger. The reason the capacitor in this application must be so large is that the capacitor must hold enough charge to supply current into the load for the much longer duration (milliseconds) of an audio signal peak, rather than the short duration (microseconds) of a class-D switching period. Again, this is not a useful solution for portable devices.
These issues limit the power available to drive a loudspeaker or other load device, particularly in mobile applications.